Reconstitution of the Receptor for Immunoglobulin E into Liposomes REINCORPORATION OF PURIFIED RECEPTORS*

Mast cells and related cells have on their surface receptors that bind immunoglobulin E (IgE) with high affinity and which, when aggregated, trigger exocy- tosis. We recently demonstrated that when these receptors are solubilized with mild detergents, their subunits dissociate unless an appropriate 1ipid:detergent ratio is maintained. The conditions required to maintain the receptors’ integrity appeared to parallel those previ- ously determined as necessary to obtain adequate incorporation of unpurified IgE-receptor complexes from detergent extracts into liposomes. We now show that purified IgE-receptor complexes having the full complement of subunits become preferentially inserted into liposomes. If the receptor subunits are chemically cross-linked to each other, at least some of such receptors can be incorporated, even though lipid is omitted during their purification. The findings suggest that the IgE-binding CY subunit of the receptor is anchored to the bilayer by means of one or both of the other subunits. in a Other Methods-Iodination of partially purified IgE-receptor com- plexes was performed as described previously (4). For cross-linking studies, the reducible hifunctional reagent dimethyl 3,3'-dithiohispro- pionimidate dihydrochloride (Pierce Chemical Co.) was used (10). Immunoprecipitations and analyses on polyacrylamide gels were per- formed essentially as described previously (4).

The initial event in the degranulation of mast cells and basophils that is mediated by IgE' involves aggregation of the receptors for IgE on the plasma membrane of such cells (1). In order to study the early reactions associated with activation under well defined conditions, we are attempting to prepare liposomes that indvidually contain multiple copies of native, purified receptors. We have empioyed a tumor analog of normal cells, an RBL cell line, as the source for the receptors ( 2 ) .
Our previous studies (3) using unfractionated detergent extracts of RBL cells showed that the efficiency of incorporation of receptors into liposomes varied when different detergents and lipids were used. Furthermore, the extent of incorporation was sensitive to the micellar ratio of detergent to phospholipid, designated p (3). In parallel studies, we found that in order to preserve the subunit structure of the receptor, conditions similar to those found to be effective for its incorporation were required (4). In that study, we observed two new components of 45 and 20 kDa, respectively, in addition to the previously described CY and subunits. Subsequent * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ Present address, Weizmann Institute of Science, Rehovot, Israel. 1  analysis has shown that the 20-kDacomponent is a disulfidelinked dimer of a 10-kDa polypeptide that has the properties of an additional subunit ( y ) of the receptor (5). The 45-kDa component proved to be a complex that upon reduction yielded intact 4 and y chains and that appears to be variably generated by aggregation of the receptors (6).
In the present work, we have attempted to incorporate purified receptors into liposomes and have examined the subunit composition of the material that became inserted.
Lipids-Phospholipids were prepared from frozen RBL tumors or rabbit livers and quantitated as reported previously (3). a-L-Phosphotidylcholine (lecithin) type IV-S from soybean was purchased from Sigma and was used without further purification.
Preparation of IgE-Receptor Complexes-Receptors were purified from the 2H3subline (13) of RBL cells. Tumors were grown for 6-8 days in WKY-Nweanling rats by injecting lo7 washed, tissue cultured cells intraperitoneally (12), and a membrane preparation was obtained as follows.* The minced tumors were homogenized in the presence of 1 ggjml of leupeptin (Boehringer-Mannheim),3 gg/ml of pepstatin A (Sigma), and 0.5-0.7 trypsin inhibitor units/ml of aprotinin (Sigma) in a Dounce homogenizer (A pestle) by three to four strokes in isotonic sucrose/Tris solution (0.25 M sucrose and 10 mM Tris-HCI, pH 7.4) followed by three more strokes with a B pestle. The homogenate was centrifuged for 5 min a t 12,000 X gav, and the supernatant microsomal fraction was then pelleted by centrifugat,ion for 70 min at 48,000 X gav. The pellet was resuspended, washed, and used within 24 h. The receptors were saturated with IgE a t 4 "C and the microsomes were washed by three 1-h centrifugations in an SW 50.1 rotor, through 5% Ficoll in sucrose/Trisat 45,000 rpm in an L5-50 or L8-70 ultracentrifuge (Beckman). Thepellets were homogenized briefly between and after the centrifugations.
Cultured RBL cells were prepared as described (14). Cells were routinely used 10 days after transfer, with one change of medium 2-3 days prior to harvesting. Cells were harvested with trypsin (M. A. Bioproducts, Walkersville, MD), washed twice in Tris/saline (15 mM Tris-HCI, pH 7.4, 124 mM NaC1, and 5 mM KCL), saturated with lgE, and used after washing as such or after preparation of a crude membrane fraction following sonication under hypotonic conditions Membranes or cells were solubilized a t a micellar ratio of detergent: phospholipids of 1.5 5 p 5 2.5 by mixing a preparation containing 2-8 mM phospholipid with appropriate amounts of the zwitterionic detergent CHAPS and incubating the mixture 1 h on ice. Soluble receptors were recovered from the supernatants after centrifugation for 1 h at 31,000 X gav. The parameter p was calculated using 4.7 mM for the cmcelfectlve of the detergent (3) i.e. p = ([detergent] -4.7)/ Iphospholipid] where the concentrations are millimolar.
Phospholipid phosphate was measured in membranes prepared from tumors or broken cells; an approximation of 2 mM phospholipid for 5 X 10' cells/ml was used for suspensions of intact cells (3).
Purification of receptors to which arsonylated IgE was hound was performed on affinity columns conjugated with purified anti-benzenearsonate antibodies (12). Partial purification of receptors bearing arsonylated anti-dinitrophenyl mouse IgE was performed on columns o f Sepharose conjugated with trinitrophenyllysine (10).
Incorporation of Receptors into Liposomes-Incorporation of receptors was achieved by raising the lipid concentration in the preparation of purified receptors with presonicated liposomes to 8-10 mM phospholipid, adding detergent to reach the desired value of p , and incuhating for 1-2 h on ice. Mixtures were then dialyzed against Tris buffer containing 1 mM NaNs for 2 days with two to three changes of buffer/day at 8-12-h intervals. The original and first two changes contained 3% sucrose.
Two types of density gradients were used. 1) A 5.3-1111 gradient of :3-36% sucrose was overlaid with 0.2-1 ml of sample and centrifuged 24 h in an SW 50.0 rotor a t 45,000 rpm at 4 "C in a Beckman L5-50 or L8-70 ultracentrifuge. Fractions of three drops were collected from the bottom (20-23 fractions/gradient) and counted for radioactivity in Reckman 7 counters (Model 8000 or Biogamma). Fractions were sometimes assayed for refractive index.
2) Stepwise gradients were constructed in Airfuge (Beckman) test tubes as described (15). These gradients contained 40 pl of 27%, 35 pl of 25%, 30 pI of 20%, 25 pl of 5'i. and 20 p1 of 0% sucrose. The specimen (30 pI in 30% sucrose) was injected under the gradient and the tubes centrifuged for 30 min in a cold room with 30 p.s.i. of air pressure. Fractions of 25 pl were collected from the top, and the same fractions from two to three replicate gradients were pooled for further processing.

RESULTS
Receptors complexed with arsonylated IgE were partially purified maintaining a detergent:phospholipid micellar ratio of -2. They were iodinated and then divided into two portions. Both were further purified on columns conjugated with antibenzenearsonate antibodies, one in the presence of phospholipids, the other with a solvent containing detergent only. Fig. 1 shows an autoradiogram of a polyacrylamide gel on which the purified receptors were analyzed. The material isolated in the absence of lipids (lane I) shows the IgE near the origin (as well as a high molecular weight breakdown product of IgE) and a weakly iodinated (Y chain appearing as a diffuse band at -60 kDa. Except for some material running ahead of the buffer front (likely representing labeled lipids), only traces of additional bands were seen. On the other hand, the material isolated in the presence of phospholipids (lane 2 ) shows a prominent doublet a t 31 and 34 kDa, representing the 8 subunit, and the disulfide-linked dimer of y chains a t 20 kDa ( 5 ) .
The patterns obtained by immunoprecipitating the eluates with anti-IgE (lanes 3 and 4 ) are virtually identical with those seen by examining the eluates directly. The immunoprecipitates from both eluates contain the (Y chains along with the IgE, indicating that the former are still bound to the latter. With the material containing the B and y chains, these are also precipitated, demonstrating that they are still associated with the (Y chains. A quantitative analysis of the material in lanes 3 and 4 is given in Table I. We have described elsewhere that upon immunoprecipitation, a 45-kDa component representing a disulfide-linked complex of / 3 and y chains is variably generated (6). As can be seen in Fig. 1, none of this component appeared in these immunoprecipitates although a subsequent immunoprecipitate of the same material did (below).

FIG. 1. Analysis of purified IgE-receptor complexes by polyacrylamide gel electrophoresis in Na+ dodecyl sulfate.
After saturation of the receptors with '"I-IgE, membrane fractions of RBL cells were solubilized at p = 2.5 (9.5 mM endogeneous lipids, 28.4 mM detergent). The IgE-receptor complexes were partially purified a t p = 2, iodinated with "'1. and then purified further on antibenzenearsonate columns in the presence (lanes 2 and 4 ) or absence (lanes I and 3 ) of 2 mM tumor-derived phospholipids along with 10 mM detergent. The column eluates were examined directly (lanes I and 2) and after immunoprecipitation with anti-IgE (lanes 3 and 4 ) .
The specimens were dissolved in Na' dodecyl sulfate without reducing agent. The figure is an autoradiogram of the 12.5% gel. Lanes I and 2 contained 850 cpm and hnes 3 and 4, i60 cpm of '"'I-labeled IgE. or., origin; df., dye front.

TARIX I
Analysis of immunc~prrcipitates of IgE-rrcrptor complexes The complexes were analyzed on polyacrylamide gels. Specimens 1 and 2 are illustrated in Flg. 1, lanrs 3 and 1, respectively, and specimens 3 and 4 in Fig. 3 , lanes I and 2, Fig. 2. Only the IgE is labeled with "'1 so that the position of these counts (filled circles) indicates the distribution of the IgE. Both the IgE and the receptors are labeled with I2'I, the IgE much more so than the receptor (Table I) In order to examine the incorporated material more completely, some of the material analyzed in lane 2 of Fig. 1, i.e. intact receptors, was reconstituted and the liposomes applied to a sucrose gradient in amounts sufficient for subsequent analysis on gels. The gradient pattern showed "'1 counts (on IgE) at both the position where free proteins travel ( 7 = 1.38) and at the position of the liposomes ( 7 = 1.347) in about equal amounts. It is likely that during the week that intervened between the first and second study, some dissociation of the IgE from the receptor, or of the subunits of the receptor from each other, or both, had occurred. Material from the top and bottom of the gradient was solubilized, immunoprecipitated, and analyzed on a polyacrylamide gel. An autoradiograph of the gel is shown in Fig. 3. The material from the lower portion of the gradient (lane 1 ) showed only the band of IgE and a very faint band at the position of the CY chains (not seen in the reproduction). The material incorporated into liposomes, on the other hand, showed the fi and y components as well (Fig. 3, lane 2). In addition, a prominent band at 43 kDa is seen. As already mentioned, this material represents disulfidelinked p-y complexes that form variably (cf. Fig. 1, lane 4 ) after immunoprecipitation (6).
A quantitative analysis of this gel is also presented in Table  I. Together, the results confirm that the receptors incorporated into the liposomes (specimen 4) have the same subunit composition as those freshly isolated (specimen 2). On the other hand, the IgE-cu chain complexes that are not incorporated into liposomes (specimen 3) are largely free of p and y chains.
Cross-linked Receptors-In order to obtain further information on the conditions required for incorporation of the receptors into lipsomes, we performed experiments on receptors that had been stabilized by chemically cross-linking them prior to purification. The following experiment is one of several performed, each of which gave similar results.
IgE-receptor complexes were partially purified from membrane preparations of cultured cells and extrinsically iodinated. An aliquot was chemically cross-linked with a reducible bifunctional reagent (see "Materials and Methods"). Both the cross-linked and uncross-linked preparations were then further purified on anti-benzenearsonate columns in the presence or absence of lipids. Fig. 4 shows an analysis on polyacrylamide gels of the four eluates. of y chains, respectively. In addition, a band at -10 kDa is seen, likely representing reduced y chains. The uncross-linked preparation that had been exposed to lipid-free detergent solutions shows, as expected, little of the /3 and y components (lane 3 ) . Lanes 1 and 2 show the material that had been crosslinked and subsequently purified in the absence (lane 1 ) or presence (lane 2 ) of lipids. In addition to the IgE at the origin, the principal component is a band at 90-95 kDa in both specimens. Traces of material with an apparent molecular weight between that of the cross-linked material and the a chains are also seen. Fig. 4B shows an autoradiogram of an 8% acrylamide gel on which the same specimens were analyzed. The intermediate sized material can be better appreciated on this gel, and the apparent molecular weight of the major cross-linked product more precisely assessed. The cross-linked product from Fig. 4 A , lane 1, was excised and re-electrophoresed on a gel, along with material excised from the bands at 30, 20, and 10 kDa from the specimen shown in lane 4 of the same gel. All the specimens were reduced. The cross-linked material yielded bands a t -30 and 10 kDa and a faint band a t 60 kDa; the band at 30 kDa isolated from the specimen in lane 4 remained unchanged and the 20-and 10-kDa components from this specimen both ran as 10-kDa components (Fig. 5). These results demonstrate that the cross-linked material contains a, /3, and y chains.
The ability of the IgE-receptor complexes to incorporate into liposomes was assessed, the samples being analyzed by centrifugation in an Airfuge (15). Fig. 6 shows an analysis on polyacrylamide gels of the fractions from the top of the The 95-kDa band seen with cross-linked receptors (Fig. 4, lane I ) was excised, reduced with 1 mM dithiothreitol, and reanalyzed on a 14% gel (lane 1). The bands a t 30, 20, and 10 kDa (Fig. 4, lane 4 ) were similarly analyzed on lanes 2.3, and 4, respectively. stepwise gradients (7') representing liposomes and from the unincorporated material at the bottom ( B ) . The material that had not been cross-linked and that had been exposed to lipidfree detergents incorporated poorly. On the other hand, the uncross-linked material that had not been exposed to lipidfree detergents and both of the cross-linked preparations showed more substantial and roughly equivalent incorporation into the liposomes.

DISCUSSION
Several years ago, we observed that when labeled in situ with the hydrophobic probe iodonaphthylazide (16), the 8 subunit but not the a chain of the receptor for IgE became modified. We suggested that possibly the a chains were anchored into the plasma membranevia the /3 subunit (17). We subsequently observed that in order to incorporate IgE-receptor complexes into liposomes efficiently, low ratios of 1ipid:detergent had to be maintained. This in turn suggested that similar conditions would be useful in overcoming the unusual instability of the subunit interactions of the receptor (9), and this proved to be the case (4). Coincidently, an  Fig. 4, were subjected to the protocol for reincorporation. In this instance, soybean lecithin was used as the source of lipids, and the original mixtures contained 10 mM lipids, 23.7 mM detergent ( p = 1.9). After dialysis, the samples were applied to stepwise gradients and centrifuged in an Airfuge (see "Materials and Methods"). Eight fractions were collected from each gradient. Fraction 1 from the top containing liposomes ( 7 1  additional pair of subunits (y chains) was uncovered when the "permissive" conditions were used during purification of the receptor ( 4 5 ) . In the present studies, we have shown that receptors purified in a way such that the @ and y chains are largely dissociated incorporate poorly into liposomes. Contrariwise, receptors that have the full complement of subunits, either because they were purified in the presence of the permissive 1ipid:detergent concentration ratios or because they were stabilized by chemical cross-linking, incorporate more effectively.
The results with the cross-linking reagents were informative with respect to the structure of the receptor. As can be seen best in Fig. 4R, the largest cross-linked species has an apparent mass of 95 kDa (the narrow band at -115 kDa is a breakdown product of the IgE (see Fig. 1 and Ref. 18)). This 95-kDa estimate corresponds closely to what would be expected if the subunits, of which the receptor is composed, were cross-linked to each other a t a molar ratio of 1 a:l @:2 y. Independent data have suggested that this is the composition of the native receptor ( 5 ) . That there is an apparent parallel conversion of the uncross-linked CY, @, and y components when the receptor is cross-linked (cf. Fig. 4A, lanes 2  and 4) is also consistent with such a model. It may be of interest to examine more closely the partially cross-linked species seen best in Fig. 4B, lanes 1 and 2. Such species may be useful for defining the sites of interaction between the subunits.
Returning to the results of the incorporation studies, several questions remain. Our experiments cannot exclude a model that postulates changes in the conformation of the n chains rather than the dissociation of the @ and y chains per se to explain the ineffective incorporation of IgE-n chain complexes. In such a model, the protective effect of chemical cross-linking would result from stabilization of the conformation of (Y chains rather than from maintenance of the interactions between the n, 8, and y subunits. If this is not the case and the other subunits are required for anchoring N, which of the former play the primary role? Since like the 8 chains, the y chains are modified by the iodonaphthyl reagent ( 5 ) and by other criteria show similar hydrophobic properties: we cannot yet determine the relative contributions of the different subunits to the effectiveness of incorporation.
These questions will be difficult to resolve and seem to us of second order importance. More significant questions are those related to the functioning of the receptor. The results presented in this paper will, we think, provide a useful step in exploring this aspect. However, further modifications of our procedures will be necessary in order to incorporate multiple copies of receptors into individual liposomes (or bilayers). In some cases, substantially nonrandom ( i e . skewed) distributions of incorporation into liposomes have been observed (19,20). If a similar phenomenon were to pertain with respect to the receptor for IgE, we would be even closer to our goal than the present results suggest.
Recent studies on receptors for IgG on macrophages (21, 22) and on the receptors for IgE (23, 24) have implicated the formation of ion channels as a direct consequence of the aggregation of the receptors. Our progress in maintaining the native structure of the receptor and in reincorporating the purified material into liposomes will allow us to probe this mechanism.